Presentation 2014-03-14
Online Failure Prediction with Accurate Failure Localization in Cloud Infrastructures
Hiroshi OTSUKA, Kaustubh Joshi, Matti Hiltunen, Scott Daniels, Yasuhide MATSUMOTO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The ability to predict the occurrence of a failure in a large scale IT infrastructure, such as a cloud data center, opens up possibilities to alleviate or completely hide the impact of the failure by taking corrective actions before the failure actually occurs. For example, a VM can be migrated from a failing physical machine to another. However, existing failure prediction methods often do not provide enough information about the location of the failure (e.g., which physical machine) to take meaningful actions. We propose an integrated method that combines the prediction of the occurrence of a failure (of a given type) with the prediction of the location of the failure enabling automated or operator actions. We describe the approach and show initial results in a commercial cloud infrastructure.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Failure Prediction / Failure Diagnosis / Failure Localization
Paper # SC2013-20
Date of Issue

Conference Information
Committee SC
Conference Date 2014/3/7(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Services Computing (SC)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Online Failure Prediction with Accurate Failure Localization in Cloud Infrastructures
Sub Title (in English)
Keyword(1) Failure Prediction
Keyword(2) Failure Diagnosis
Keyword(3) Failure Localization
1st Author's Name Hiroshi OTSUKA
1st Author's Affiliation System Software Laboratory, FUJITSU LABORATORIES LTD.()
2nd Author's Name Kaustubh Joshi
2nd Author's Affiliation AT&T Labs-Research
3rd Author's Name Matti Hiltunen
3rd Author's Affiliation AT&T Labs-Research
4th Author's Name Scott Daniels
4th Author's Affiliation AT&T Labs-Research
5th Author's Name Yasuhide MATSUMOTO
5th Author's Affiliation AT&T Labs-Research
Date 2014-03-14
Paper # SC2013-20
Volume (vol) vol.113
Number (no) 496
Page pp.pp.-
#Pages 6
Date of Issue